ECG Arrhythmia Classification Using Ensemble of Features

Abstract

Cardiovascular diseases are one of the most common cause of fatality across the world. This work aims to develop a cognitive support system which can aid in detection and classification of multiple arrhythmias, namely premature ventricular contraction (PVC), right bundle branch block (RBBB), left bundle branch block (LBBB) and paced(P) have been studied. An ensemble of features has been created using time domain features, statistical features, and entropy-based features. The ensemble feature vector is used to train the multi-layer perceptron. Experiments have been performed on 24 subjects of the MIT-BIH arrhythmia dataset. The classification accuracy, precision, sensitivity, and specificity were computed for the proposed scheme, and the results obtained outperformed those obtained with a single set of features.